Random Subspace Aggregation for Cancer Prediction with Gene Expression Profiles

Table 5

Testing error comparison of RS_SVM and the state-of-the-art methods (%).

	Breast Cancer	Leukemia	Lung Cancer	Prostate	Colon Tumor	CNS	Ovarian	DLBCL
RS_SVM	5.30	5.89	1.34	0	14.52	33.33	1.19	4.26
Nanni et al. [28]	11.43	0	0	3.85	26.67	33.33	0	1.43
Ye et al. [29]	—	2.50	—	7.5	15.00	—	—	—
Liu et al. [30]	—	0	0	3.00	8.10	—	0.80	2
Tan and Gilbert [31]	—	8.90	6.80	26.50	4.90	11.7	—	—
Ding and Peng [32]	—	0	2.70	—	6.50	—	—	—
Bonilla Huerta et al. [33]	—	0	0.70	4.00	8.1	13.40	0	0
Cheng [34]	—	0	0.67	5.88	—	—	—	—
Paliwal and Sharma [35]	26.3	0	2.70	23.5	—	—	—	—
Bolón-Canedo et al. [10]	36.22	11.96	2.75	11.81	13.10	36.67	1.20	20.50
	46.56	4.11	0	41.87	16.19	30.00	0.8	6.50
	28.11	5.54	1.11	12.53	19.05	36.67	0	4.00
Porto-Díaz et al. [36]	21.05	0	0.67	20.59	10.00	25.00	0	0
Hu et al. [37]	—	—	12.50	19.30	9.70	—	—	—
	—	—	11.60	18.20	9.70	—	—	—
Nagi and Bhattacharyya [11]	26.51	7.55	18.12	47.06	5.60	9.85	1.11
Pati and Das [38]	—	7.89	6.25	—	—	—	—	—
Dash et al. [39]	—	0	11.55	—	10.95	—	—	—
	—	0.45	0	—	0	—	—	—
	—	28.22	16	—	23.33	—	—	—
	—	0.41	0.95	—	0.31	—	—	—
Ghorai et al. [12]	18.79	5.48	3.62	9.84	17.23	—	—	—
Luo et al. [13]	—	2.07	—	—	18.60	—	—	6.00
	—	2.45	—	—	19.12	—	—	7.19

The state-of-the-art methods are indexed by the first author in literatures. “—” means that there are no corresponding results in the given literature.